Mojo function
preferred_simd_width
preferred_simd_width[dtype: DType]() -> Int
SIMD scalar count for fused GPU concat vectorization.
Uses 32-byte global loads on sm_100a; otherwise the target's native
simd_width_of for dtype on the active GPU compilation target.
Returns:
Was this page helpful?
Thank you! We'll create more content like this.
Thank you for helping us improve!